面向高精度航天器状态估计系统的预测变结构滤波理论研究

High-performance filtering technology is the theoretical basis and core for the state estimation of spacecraft with high-precision; therefore, the filtering theory research for high-precision, robustness and low calculation burden is the key to improve the technology level of spacecraft state estimation. However, due to the particularity of space missions, the traditional filter theories are difficult to meet the above technical requirements with the existence of the strong nonlinearity and uncertainty. Hence, this project will focus on a new type of filtering technology——Nonlinear Predictive Variable Structure Filter with in-depth research and exploration. Due to the filter has the advantages of stable precision, low calculation burden and don’t need to satisfy the Gaussian white noise assumption, which is a kind of ideal nonlinear filtering method. Considering the theoretical requirements and technical characteristics of high-precision spacecraft, the three aspects are chosen for the breakthrough of theoretical research, such as algorithm optimal design, robust research and data fusion. As a result, a series of filter methods are put forward to meet different demands of the spacecraft. By discussing the applications in the high precision spacecraft, the feasibility and effectiveness will be verified for the proposed filter methods. In addition, through the research in this project will establish a systematic, complete theoretical system for the predictive variable structure filter, in order to provide a theoretical basis for its popularization and application in other fields.

高性能的滤波技术是高精度航天器状态估计技术的理论基础与核心，因此，针对滤波精度高、鲁棒性能强且计算量有限的滤波理论研究是提高航天器状态估计水平的关键所在。但是，由于航天任务的特殊性，航天器系统往往具有较强的非线性和不确定性，这就导致传统的滤波方法难以满足上述要求。为此，本项目将对一种新型滤波器——非线性预测变结构滤波器进行深入研究，该滤波器具有精度稳定，计算量小且无需满足高斯白噪声假设等优点，是一种理想的非线性滤波方法。结合高精度航天器的理论需求与技术特点，从高精度算法优化设计、鲁棒性能研究和数据融合三个方面入手重点突破，提出了一系列满足不同航天任务需求的滤波方法。通过探讨在高精度航天器状态估计中的应用，验证所提出理论方法的可行性和有效性。此外，本项目的研究将建立较为系统、完善的预测变结构滤波理论体系，为在其他领域的推广应用提供较好的理论基础。

由于航天器动力学难以精确建模，导致其高精度的导航系统设计必须依赖于高精度、高可靠度的导航滤波理论方法。现有的导航滤波理论方法主要是基于Kalman滤波的方法理论体系，对于存在不确定性模型误差的航天器系统而言，难以实现高精度导航状态估计。本项目通过对预测滤波及预测变结构滤波的深入理论研究，提出了多种抑制不确定性模型误差的理论方法，具体为：研究了一类非线性自适应预测变结构滤波方法，增强了对模型误差复杂多变系统的理论适应性；研究了一类Sigma点预测滤波及预测变结构滤波方法，可以有效解决预测滤波加权矩阵难以选择的问题，提升了预测滤波及预测变结构滤波的理论估计精度；研究了一类鲁棒滤波理论方法，可以有效提升预测变结构滤波对于各种噪声及模型误差的抑制能力，具有广泛应用价值；研究了将滤波估计理论方法融入高精度控制器设计的策略，提升了传统控制器对于模型误差的补偿能力，进而提升整体控制精度。最后，将所提的设计方法应用到航天器姿态确定与控制系统的设计中，开展满足实际工程需求的应用研究。